Null string is encoded as "null" in incremental index #2765
Conversation
|
Is it possible to include a test that would have caught this? |
| { | ||
| return String.valueOf(o); | ||
| return o == null ? null : String.valueOf(o); |
There was a problem hiding this comment.
can you explain in your PR description why this line is not enough to fix the problem. What is the purpose of the remaining changes?
|
Can we remove the changes not related to the String null handling? The code being refactored is being redone in #2760, e.g. those value transformers in IncrementalIndex are removed. |
|
@jon-wei sure |
navis
force-pushed
the
invalid-encode-nullstring
branch
from
0f5a845
to
f0e55f5
Compare
|
Is this a bug that affects 0.9.0, or just 0.9.1? |
|
@drcrallen it's from #2263, which is not included in any released versions. |
|
@drcrallen it'll just be for 0.9.1, the code being changed was from #2263, which isn't in 0.9.0 |
|
👍 |
| @@ -101,7 +101,7 @@ | |||
| @Override | |||
| public String apply(final Object o) | |||
| { | |||
| return String.valueOf(o); | |||
| return o == null ? null : String.valueOf(o); | |||
There was a problem hiding this comment.
Should this be converting "" to null or is something else doing that?
There was a problem hiding this comment.
I think it's fine, null/empty dim values will end up as "" via a nullToEmpty() call when added to the dictionary:
There was a problem hiding this comment.
ah ok, this looks good then. thanks
|
👍 |
| try { | ||
| return Long.valueOf((String) o); | ||
| return s.isEmpty() ? null : Long.valueOf(s); |
There was a problem hiding this comment.
does this change any current behavior that might map empty strings to 0 for numeric values?
There was a problem hiding this comment.
I don't think it could, since with the old code, Long.valueOf("") would have thrown an exception. The new code would return null.
I've found null values for string type is encoded as "null", which seemed not intended behavior.